News Chat - Structured Current Event Intelligence

Replaces noisy web-search retrieval with deduplicated, clustered, and entity-aware event records.
Improves response quality for timelines, location-specific updates, and topic summaries.
Combines fast retrieval, event grouping, and filtering by themes, people, organizations, and locations.
Uses an MCP server integrated with GDELT to give AI agents structured access to current event data.

For early access to integrate the News MCP server, please contact us.

A news assistant built on structured event data, not overlapping search results.

Structured Event Retrieval

The system retrieves current events from GDELT through an MCP server, giving agents access to structured records instead of raw search-result pages.

That enables more precise filtering, better chronology, and stronger grounding for answers about recent developments.

Deduplication and Clustering

Similar articles are deduplicated and then clustered so multiple reports about the same event are treated as one coherent event group.

This reduces redundancy and gives the AI a clearer understanding of the event itself rather than isolated documents.

Entity and Theme Filtering

Events can be filtered using themes, people, organizations, and locations extracted from the source data.

The result is more relevant answers for location-specific updates, subject summaries, and targeted monitoring use cases.

How News Chat Works

Data Ingestion

Pulls fresh GDELT event files every 15 minutes.
Parses article metadata, entities, and themes.
Stores structured records for downstream retrieval.

Data Processing

Removes duplicates using embedding similarity.
Clusters related articles into event groups.
Aggregates records into broader domains and topics.

MCP Retrieval Layer

Exposes search tools for clusters and entities.
Returns structured JSON for agents to analyze.
Supports graphs, timelines, and related-event lookups.

AI Response

Generates cleaner summaries of current events.
Builds timelines and location-based updates.
Responds with broader coverage and less repetition.

Best-Fit Use Cases

News Chat is designed for questions about latest events, chronological timelines, entity-specific developments, and category-grouped summaries for a location or topic.

Why It Improves Accuracy

By giving the agent clustered event records instead of overlapping web pages, the system improves coverage while reducing duplicate evidence in the final answer.

Summarize the latest events in a specific location and group them by topic.

News Chat retrieves clustered event records, filters by location and themes, and returns a cleaner topic-based summary with stronger chronology and less duplication.